Anais do XIX Simpósio Brasileiro de Sensoriamento Remoto

A sandbank consists of the accumulation of sediments (sand and gravel) deposited in a riverbed or along the coast, constituting an obstacle to navigation. This work aimed to develop a technique for the detection of sandbanks located in an Amazonian estuarine region (Guajará’s bay) using synthetic aperture radar images time series in order to assist the obstacles monitoring in the local navigation. We used 28 images of the Sentinel-1 Mission; Orbital Platform: S1A; Band: C; Product Type: Ground Range Detected (GRDH); Polarizations: VV and VH; Sensor Mode: Interferometric Wide Swath Mode (IW). The results showed a significant stratification for 3 classes (Water, Sandbank and River Island), with accuracy equal to 91,1% for an automatic classification by the k-NN algorithm.